# Bull/bear Consensus and Puetz Windows, from Jeff Sasmor

May 29, 2010 |

I wonder if the computers that do most of the trading have programmed Puetz windows, bull/bear ratios, etc. into their algos. Mostly they seem to trade off newsflow and momentum. They seem to find attractors in moving averages, pivots points, and big point numbers.

My own experience and training as an EE can only compare the current environment to a closed-loop feedback control system gone open-loop; as if elephants constrained in their pen in the zoo have burst out of their cages, running this way and that as they bang into walls.

## Phil McDonnell writes:

The trouble with the Puetz window is that an eclipse sounds like a very rare occurrence. After all I have only seen maybe 1 or 2 eclipses in my lifetime. Would you be surprised to know that 4 of them occur every year and sometimes more than 4. There are 2 solar and 2 lunar ecipses every year minimum. The fact is that they only occur in certain relatively small swaths somewhere on the Earth. So the 4 that will occur this year will probably not be visible to most people. Add to that the fact that we are less likely to go outside at night or be asleep explains why we see them so rarely.

Next consider the Puetz window definition. Start with the 4 eclipses then construct a 12 week window from 6 weeks before to 6 afterwards. That covers a period of 12 x 7 x 4 = 336 days if eclipses were randomly spaced. Adjusted for the relationship between solar and lunar eclipses the the Puetz window covers 196 days out of the year. That is more than half of all the days in a year. Adding the full Moon cuts down the number of allowable days to 10 days, 6 before, 3 after and the day of the full Moon. The allowable number of days in the window every year now becomes an average of 68 or 18% of the year.

This has all the classic earmarks of a theory that was hand fitted to a small sample of 8 crashes.

`SELECT * FROM wp_comments WHERE comment_post_ID = '4806' AND comment_approved = '1' ORDER BY comment_date`